topic How to code the recurrent events in SAS Programming

How to code the recurrent events

SeaMoon_168 — Thu, 15 Aug 2024 21:40:52 GMT

I have a dataset that have four variables: ID, Day, Status, and Var1.

ID	Day	Status	Var 1
1	1	0	1
2	1284	0	2
3	28	0	2
4	432	1	1
5	1018	0	2
6	85	0	1
7	1007	0	2
8	824	0	2
9	907	0	2
10	191	1	2
10	392	1	2
10	433	1	2
11	819	1	2
11	1004	1	2

However, since ID 10 and 11 have multiple events and I want to code "Day" as two variables "Start" and "End". Many thanks!

ID	Start	End	Status	Var1
1	0	1	0	1
2	0	1284	0	2
3	0	28	0	2
4	0	432	1	1
5	0	1018	0	2
6	0	85	0	1
7	0	1007	0	2
8	0	824	0	2
9	0	907	0	2
10	0	191	1	2
10	191	392	1	2
10	392	433	1	2
11	0	819	1	2
11	819	1004	1	2

Re: How to code the recurrent events

ballardw — Thu, 15 Aug 2024 22:16:10 GMT

Is your data actually sorted as shown?

Please see:

data have;
  input ID 	Day 	Status 	Var ;
datalines;
1 	1 	0 	1
2 	1284 	0 	2
3 	28 	0 	2
4 	432 	1 	1
5 	1018 	0 	2
6 	85 	0 	1
7 	1007 	0 	2
8 	824 	0 	2
9 	907 	0 	2
10 	191 	1 	2
10 	392 	1 	2
10 	433 	1 	2
11 	819 	1 	2
11 	1004 	1 	2
;


data want;
   set have;
   by id;
   l_day=lag(day);
   if first.id then do;
      start=0;
      end=day;
   end;
   else  do;
      start=l_day;
      end=day;
   end;
   drop day l_day;
run;

Note the first data step creates a usable data set as shown. This is the preferred manner of sharing data on this forum as then we know the variable names, types and any properties set such as formats or labels.

The second data set assumes the values are sorted by Id and Day. This allows use of a simple By Id. When you use a By statement in the data step then SAS creates automatic variables named First.variable and Last.Variable that have values of 1/0 indicating true/false that the current observation is the first of a by group or last of the by group.

Since you need a value from the previous observation we use the LAG function to get the value to use as needed. Warning: this function is a queuing function and seldom returns the result wanted when used conditionally (in an IF <condition> then do; <statements>; end; block ).

Then knowing whether the observation is the first or not we know which value to use for Start and End. Then drop the no longer needed variables.

Re: How to code the recurrent events

mkeintz — Fri, 16 Aug 2024 00:21:43 GMT

Assumy data are sorted by ID/DAY, then:

data have;
  input ID 	Day 	Status 	Var ;
datalines;
1 	1 	0 	1
2 	1284 	0 	2
3 	28 	0 	2
4 	432 	1 	1
5 	1018 	0 	2
6 	85 	0 	1
7 	1007 	0 	2
8 	824 	0 	2
9 	907 	0 	2
10 	191 	1 	2
10 	392 	1 	2
10 	433 	1 	2
11 	819 	1 	2
11 	1004 	1 	2
;

data want;
  set have;
  by id;
  end=day;
  start=ifn(first.id,0,lag(end));
run;

Re: How to code the recurrent events

SeaMoon_168 — Fri, 16 Aug 2024 14:05:08 GMT

Many thanks for your help. It is easily understood and works well.

Re: How to code the recurrent events

SeaMoon_168 — Fri, 16 Aug 2024 14:05:47 GMT

Thank you for your help.